Valproic acid sensitizes pancreatic cancer cells to natural killer cell-mediated lysis by upregulating MICA and MICB via the PI3K/Akt signaling pathway

Seventy-eight patients with pancreatic ductal adenocarcinoma (PDAC) underwent surgical treatment in Pancreatic Disease Institute, Union Hospital (Wuhan, China) during June 2012 and December 2012 (aged between 33 and 79; median age, 56 years; 45 males and 33 females). The surgical specimens were studied retrospectively. The samples were fixed in 4% formalin solution for 18-24 hours and embedded in paraffin for immunohistochemical analysis. The diagnosis of all patients was confirmed by histologic examination. The use of the clinical samples for analysis was approved by the Ethics Committee of Huazhong University of Science and Technology.

Sodium valproate (VPA) and interleukin-2 was obtained from Sigma-Aldrich, St. Louis, MO, USA. Bovine serum albumin (BSA) and trypsin were purchased from Amresco, Solon, OH, USA. Fetal bovine serum (FBS), donor equine serum (DES), Alpha modified eagle medium (alpha-MEM), and Dulbecco’s modified eagle medium F12 (DMEM/F12) were obtained from Hyclone, Logan, UT, USA. Lapatinib, LY294002, rabbit polyclonal antibodies against PI3KCA, Akt Rabbit mAb, Phospho-Akt (Ser473) Rabbit mAb, HER3 Rabbit mAb, Phospho-HER3 Rabbit mAb, GAPDH Rabbit mAb, and goat anti-rabbit IgG antibodies conjugated to HRP were purchased from Cell Signaling Technology, Danvers, MA, USA. Anti-NKG2D mAb was obtained from R&D, Minneapolis, MN, USA. Phycoerythrin (PE)-labeled antibodies against human MICA and MICB and mouse IgG1 isotype control antibody were obtained from Biolegend, San Diego, CA, USA. Rabbit polyclonal antibodies against MICA and MICB were obtained from Santa Cruz, Santa Cruz, CA, USA.

The human pancreatic adenocarcinoma cell lines PANC-1, MIA PaCa-2, and BxPC-3, and the human natural killer cell line NK-92 were obtained from the American Type Culture Collection (ATCC; Manassas, VA, USA). PANC-1, MIA PaCa-2 and BxPC-3 cells were cultured in DMEM/F12 containing 10% FBS. NK-92 cells were maintained in alpha-MEM containing 12.5% DES, 12.5% FBS, and 10 ng/mL interleukin-2. All cells were cultured in incubator at 37°C in a 5% CO₂ atmosphere.

PANC-1, MIA PaCa-2, and BxPC-3 cells were cultured to 80-90% confluence, trypsinized, washed twice with phosphate buffer solution (PBS), re-suspended in PBS at 1 × 10⁶ cells/100 μl, incubated with PE-anti-human MICA and MICB antibody or an isotype control antibody for 30 min, and then analyzed on a Becton Dickson LSR II flow cytometer (BD, Franklin Lakes, NJ, USA).

Total RNA was extracted from PANC-1, MIA PaCa-2, and BxPC-3 cells using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) and reverse transcribed using SuperScript VILO cDNA Synthesis Kit (Invitrogen). The expression of human epidermal growth factor receptor 2 (HER2), human epidermal growth factor receptor 3 (HER3), ataxia telangiectasia mutated kinase (ATM), ATM- and Rad3-related kinase (ATR), MICA, MICB, PI3KCA, and β-actin were quantified using the quantitative SYBR Green PCR kit (TaKaRa Bio) according to the manufacturer’s protocol. The primers used for qRT-PCR are shown in Additional file 1: Table S1.

Whole cell extracts were prepared using RIPA lysis buffer containing 1 mM PMSF, and the protein concentrations of the supernatants were determined using the BCA protein assay kit (Thermo Scientific, Rockford, IL, USA) according to the manufacturer’s protocol. Western blots were performed following standard procedures. Densitometry was performed using Image J V.1.46r (National Institute of Health).

A siRNA targeting human PI3KCA (si-PI3KCA) was purchased from Ribobio, Guangzhou, China; a scrambled siRNA was used as a negative control (NC). PANC-1 and BxPC-3 cells were plated in 24-well plates and transfected using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s instructions. The siRNA sequences are shown in Additional file 1: Table S2.

Cytolytic activity was assayed using the standard lactate dehydrogenase (LDH) release assay. The target PANC-1, MIA PaCa-2, and BxPC-3 cells were incubated with or without 1 mM VPA for 24 h, washed, NK-92 cells were added to the target cells as effector cells, and the cells were co-cultured for 4 h at 37°C. To block NKG2D on NK-92 cells, 10 μg/ml anti-NKG2D mAb or mouse IgG1 isotype control antibody were added to the NK cells 30 min before co-culture. Spontaneous release of LDH by the target cells alone was < 15% of the maximal release of LDH by target cells lysed in 1% NP-40. The experimental LDH release values were corrected by subtraction of the spontaneous LDH release values of effector cells at the same dilution. Percentage lysis was calculated as: (corrected experimental LDH release - spontaneous LDH release) / (maximum LDH release – spontaneous LDH release) × 100.

Four-week-old female NOD/SCID mice were randomly divided into four groups (n = 5 per group) for each pancreatic cancer cell lines. The mice in the control group were subcutaneously injected into the flank with 2 × 10⁶ untreated PANC-1 cells or BxPC-3 cells, and the mice in the three experimental groups (NK, NK + VPA, and NK + VPA + LY294002) were co-injected with 2 × 10⁶ PANC-1 cells or BxPC-3 cells and 1 × 10⁷ NK-92 cells, and then repeatedly injected with 1 × 10⁷ NK-92 cells at the same site every 2 days during the experiment. The NK + VPA and NK + VPA + LY294002 groups were injected with PANC-1 cells or BxPC-3 cells which had been pre-incubated with 1 mM VPA for 24 hours and were intraperitoneally injected with 500 mg/kg VPA every 2 days during the experiment; the NK + VPA + LY294002 group were also intraperitoneally injected with 25 mg/kg LY294002 every 2 days during the experiment. Tumor volume was calculated every week using the formula: length × width² × 0.5. The mice were sacrificed 4 weeks after the initial injection and the xenografts were excised and subjected to immunohistochemical analysis. All experimental protocols were approved by the Committee on the Ethics of Animal Experiments of the Union Hospital, Huazhong University of Science and Technology.

Sections (4 μm) were prepared from the paraffin-embedded human primary tumors and mouse xenograft tumors. Immunohistochemistries were performed following standard procedures. For mouse xenograft tumors, the positive cells were counted, and the percentage was calculated. For clinical specimens, MICA and MICB expression were scored semi-quantitatively on the basis of the staining intensity and percentage of positive cells. Samples with less than 20% positive cells was considered to be weak expression, while that with more than 20% positive cells was considered to be strong expression.

Data were presented as the mean ± standard deviation for flow cytometry, quantitative real-time RT-PCR, western blotting, cellular cytotoxicity assay, and xenograft assay, analyzed by t-test. Data of clinical characteristics were analyzed by Chi-square test. A significance threshold of P < 0.05 was used. Data were analyzed using SPSS v.11 statistical software (SPSS, Inc.).

Immunohistochemistry analysis revealed the MICA and MICB expression in pancreatic cancer (Additional file 2: Figure S1). The expression of MICA and MICB in pancreatic cancer was significantly correlated with late TNM stage, tumor differentiation and lymphatic invasion. There were no obvious relationship between MICA and MICB and other clinical features such as sex, age, and distant metastasis (Additional file 1: Table S3).

We first investigated the effect of VPA on NK cell-mediated kill of pancreatic cancer cells. PANC-1, MIA PaCa-2, and BxPC-3 cells were incubated with or without 1 mM VPA for 24 h. The LDH release assay demonstrated that NK-92 cells could lyse the pancreatic cancer cells; however, after incubated with 1 mM VPA for 24 hours, the lysis of PANC-1, MIA PaCa-2, and BxPC-3 cells mediated by NK-92 cells increased from 48.11% ± 8.29% to 66.22% ± 3.22%, 34.88% ± 4.09% to 53.11% ± 8.29% and 38.68% ± 4.09% to 58.81% ± 4.96% respectively at an effector/target (E/T) ratio of 20:1. The differences were statistically significant (Figure 1A). Pre-incubation of NK cells with an anti-NKG2D antibody for 30 minutes almost completely abolished the increased NK cell-mediated lysis of pancreatic cancer cells observed in VPA-treated co-cultures, indicating that the ability of VPA to promote the NK cell-mediated lysis of pancreatic cancer cells was dependent on a NKG2D/NKG2DL interaction between NK cells and pancreatic cancer cells (Figure 1B).

The NKG2DLs MICA and MICB play an important role in the NK cell-mediated lysis of cancer cells [21]; therefore, we determined the effect of VPA on the expression of MICA and MICB mRNA in the human pancreatic cancer cell lines PANC-1, MIA PaCa-2, and BxPC-3. Real-time quantitative PCR analysis revealed that treatment with 1 mM VPA for 24 hours upregulated MICA and MICB mRNA expression significantly in PANC-1, MIA PaCa-2, and BxPC-3 cells (Figure 2A). We also examined the surface expression of MICA and MICB in pancreatic cancer cells treated with or without 1 mM VPA for 24 h. Flow cytometric analysis demonstrated that VPA significantly increased the expression of MICA and MICB on the cell-surface of PANC-1, MIA PaCa-2, and BxPC-3 cells (Figure 2B).

Expression of MICA and MICB are associated with a variety of signaling pathways, including the HER2/HER3, ATM/ATR, PI3K/Akt, and Erk pathways, in different cells [17, 22‐24]. To explore the mechanism by which VPA upregulates MICA and MICB in pancreatic cancer cells, we examined the expression and activation of components of the HER2/HER3, ATM/ATR, and PI3K/Akt pathways. Real-time quantitative PCR analysis revealed that VPA upregulated HER3 and PI3KCA, and downregulated HER2 in PANC-1, MIA Paca-2, and BxPC-3 cells. Additionally, VPA downregulated ATM and ATR in PANC-1 cells, but had no significant effect on ATM and ATR in MIA PaCa-2 and BxPC-3 cells (Figure 3A). Western blotting analysis revealed that incubation with 1 mM VPA for 24 h led to a significant increase in the expression and phosphorylation of HER3 protein (Figure 3B), as well as the phosporylated Akt in all three pancreatic cancer cell lines (Figure 3C), but not the phosphorylated Erk (Additional file 3: Figure S2).

To determine whether the VPA-induced upregulation of MICA and MICB was related to activation of the HER2/HER3, PI3K/Akt, or ATM/ATR signaling pathways, PANC-1, BxPC-3, and MIA-Paca-2 cells were exposed to 1 mM VPA for 24 h in the presence or absence of 1 μM of the HER2/HER3 inhibitor lapatinib, 10 μM of the PI3K inhibitor LY294002, or 1 mM of the ATM/ATR inhibitor caffeine. Real-time quantitative RT-PCR and flow cytometric analysis demonstrated that the ability of VPA to upregulate the expression of MICA and MICB was significantly suppressed by lapatinib and LY294002, but not caffeine (Figure 4A-C). Next, we silenced PI3KCA using a siRNA in PANC-1 and BxPC-3 cells. Western blot analysis confirmed that the expression of PI3KCA was significantly reduced in PANC-1 and BxPC-3 cells 48 h after transfection of the siRNA (Figure 4D). Real-time quantitative RT-PCR and flow cytometric analysis demonstrated that the ability of VPA to upregulate the expression of MICA and MICB was significantly suppressed by transfection with PI3KCA siRNA (Figure 4E, F). Additionally, the ability of 1 mM VPA to increase the NK cell-mediated lysis of pancreatic cancer cells was significantly attenuated by knockdown of PI3KCA (Figure 4G). Although the role of PI3KCA siRNA on the expression of MICA and MICB protein was not totally compatible with its role on the NK cell-mediated lysis, the trend suggested that PI3K/Akt pathway played an important role in VPA-induced upregulation of MICA and MICB in pancreatic cancer cells.

Results showed that treatment with VPA significantly enhanced the ability of NK-92 cells on inhibiting the growth of pancreatic cancer xenograft tumors; however, the anti-tumor effect of VPA was partly attenuated by treating the mice with the PI3K inhibitor LY294002 (Figure 5A, B). Furthermore, immunohistochemical analysis revealed that VPA significantly upregulated the expression of MICA and MICB in the tumor xenografts compared to the control group and NK-92 group, while administration of LY294002 significantly attenuated the ability of VPA on upregulation of MICA and MICB expression in the tumor xenografts (Figure 5C).